That is not me, not at all, that is this guy... Even the GDR had research projects on his topic.
I wonder, if he really understands that the LLM Ollama models are something very else of what he did, because hardly any guy on YouTube explains the reasoning part of the LLM models. Instead, they are either the end of mankind's working class on apocalyptic levels or just advanced chat systems, with some guys trying to figure out what you can do with these thingies... actually for real.
This is because of him mixing up Neural Network with Machine Learning and AI, for example. One is a designe of the other and the other does not have to be an AI. In a world in which no one differs between between Rule Based System and AI that is normal and wording depends a lot with whom someone talks.
He appears to have sever understanding of the last 25 years of Machine Learning Science. Picture recognition on an understanding level is slightly different from the LLM models that derive from the Touring Test. The Touring Test passing is about creating a chat that a human considers a human and no software. It turned out that those being more intelligent appearing are more often taken for a human.
Understanding pictures is classic machine learning science and heavily mathematics based.
The LLM and Ollama guys overtook these guys there by having kinda accidentally found NOS for their V8....
The creation of LLM systems (Large Language Models) is based on acknowledging that some words make no sense following up other words.
oxymoron
/ˌɒksɪˈmɔːrɒn/
noun
noun: oxymoron; plural noun: oxymorons
a figure of speech in which apparently contradictory terms appear in conjunction (e.g. faith unfaithful kept him falsely true ).
While none of these OGs ever considered 1+1=5 an any good joke.
That means, that the pattern recognition was solved by the developers of Ollama LLM models over trying to make chat output senseful, while he and his Gang tried to create algorithms. One is connecting words, the other using pure numbers.
He basically says that the Word Vector Database principle of LLM AI systems is so powerful that going away from mathematical research on computational intelligence towards how to train reasoning systems is the way ahead.
This being said, is a break through in algorithmic AI another needed and important step of scientific research within the IT field. Such AI is needed in TCP/IP package routing, advanced navigation, CAD and mathematical simulations
like here.
But try to explaine this as a Super Nerd the Finance Manager Guy... only, being hardcore cool and nerd is even worse: Cyber Punk! Broke Nerd Inc. They all think you live on air and your ego.
#gfyALL
#cyberpunkcoltoure
